Defining the McClellan Oscillator - Part II

This is a multi-part series poking around the edges of the McClellan Oscillator (MO), a breadth indicator measuring the degree of participation in market moves (read more).

This post isn’t about a trading strategy (yet). In this post I want to try to understand in general terms what the MO says about the market by looking at how the market has responded to (a) an increasing/decreasing MO and (b) a MO reading above/below zero. In a follow up post we’ll look at the MO at extremes.

Relationship #1: An Increasing/Decreasing McClellan Oscillator

[logarithmically-scaled]

The graph above shows the results of two strategies, one going long the S&P 500 at Tuesday’s close if the MO increased Tuesday (green) and the other if it decreased (red), from 1950. This is a proof of concept, so these results are frictionless (i.e. do not account for transaction costs or slippage).

Generally speaking, an increasing MO would indicate more stocks are advancing in recent history than more distant, or put another way, that market breadth is growing stronger.

As the graph shows, for most of the market’s history, an increasing MO (green) has been very short-term bullish, and vice-versa. But roughly around the turn of the millennium, that relationship began to reverse to contrarian. The next graph shows the same test, but only covers from 2000.

[logarithmically-scaled]

The relationship today isn’t as clear as it has been historically, but generally speaking, an increasing MO (green) is now bearish.

I think the “why” here is very simple. Days when the MO increases tend to also be days when the S&P 500 closes up, and vice-versa (because there tend to be more advancing than declining issues when the market gains). And as we’ve talked about ad infinitum on this blog (ex. here and here), at this moment in history, 'up closes' in the market tend to be followed by 'down closes', and vice-versa. In other words, the daily mean-reversion predicted by the MO is more or less the same as the daily mean-reversion predicted by daily follow-through.

Relationship #2: McClellan Oscillator Above/Below Zero

[logarithmically-scaled]

The graph above shows the results of two more trading strategies, one going long the S&P 500 at Tuesday’s close if the MO closed above zero (green) and the other if it closed below zero (red), from 1950.

Geek note: for readers who have taken the time to break down the formula used to calculate the McClellan Oscillator (read more), saying an MO above/below zero is the same as saying the 19-day EMA is above/below the 39-day EMA.

The graph shows that for most of the market’s history, higher MO readings have been more bullish than lower ones. But in the 1990’s, that relationship also reversed to contrarian. The next graph shows the same test, but only covers from 2000.

[logarithmically-scaled]

In this more recent history, MO readings above/below zero haven’t been predictive of much.

I think the “why” here is also pretty straight-forward. In MarketSci parlance (which is completely made up by me and of no use outside of this blog), the MO used this way is what I’d call an “intermediate indicator”.

And as we shown before (ex. in the moving average spectrum), intermediate-term indicators (ex. the two proprietary ones on the State of the Market) tended to flip contrarian further back in history than shorter-term ones (ex. adaptive daily follow-through).

Closing Thoughts

At first blush, the MO doesn’t tell us much beyond what price movement itself already tells us.

In upcoming posts, I’ll look at the McClellan Oscillator at extreme values (which is how it is traditionally used) and try to put my own spin on the indicator and turn it into something a bit more useful.

More to follow.

Defining the McClellan Oscillator - Part I

Comments

[LilBob:]

I agree with Cetin, this is just noise. About the only thing this whole MO concept relates to is Galbraith's idea that hyperbole is a greater predictor of stock movements than real financial numbers. You'd have better luck avoiding pseudo-mathematical models for what is inherently an ethnographic problem.

Apr 23 07:46 PM

[John Lounsbury:]

Michael - - -

Please continue to post your technical research articles. They are of great interest to some of us. I would recommend that those who do not find them useful should not be reading them.

Apr 24 01:03 PM

[Michael S:]

No sweat off John - doesn't bother me a lick. My posts get automatically reposted from my actual blog.

It's funny...my approach to the market seems to cause a visceral gag reflex in folks (particularly here at SA). If folks would just take a look at what we do, our independently-audited real-time track record, they might change their tune, but it is what it is.

Hope all is well,
michael

Apr 24 03:17 PM

[cebu sun:]

Show us what DOES work. The gag reflex is caused by your wasting time. STFU with part 1 n 2 which are zero in usefulness.

Apr 25 02:16 AM